DigitalOcean Holdings Inc. has inked a $111 million deal to acquire Paperspace Co., a Brooklyn-based startup that provides cloud infrastructure for artificial intelligence developers.

The transaction was announced this morning. 

Founded in 2011, DigitalOcean provides a public cloud platform of the same name. The platform is designed to be relatively easy to use, which has made it popular among small businesses and developers. DigitalOcean generated $165.1 million in revenue from its cloud services last quarter, up 30% year-over-year. 

Paperspace, for its part, is a venture-backed startup that raised $35 million in funding prior to the acquisition. The startup says that its user base comprises 500,000 developers who have clocked up 100 million compute hours, a measure of infrastructure usage, to date. 

Like DigitalOcean, Paperspace operates a public cloud platform. But the startup’s platform has a much narrower focus: It’s specifically built to power AI projects. Paperspace offers access to AI-optimized graphics processing units from Nvidia Corp., as well as software tools that developers can use to build and deploy neural networks.

Customers interact with Paperspace’s platform through a browser-based implementation of Jupyter Notebook. That’s an open-source coding tool widely used for AI development. One of the main reasons behind the tool’s popularity is that it allows developers to write an AI model and view the model’s processing results in the same interface, which removes the need to switch tabs.

After software teams create a neural network, they can deploy it on Paperspace’s cloud-based GPUs. The startup offers a range of compute instances equipped with Nvidia silicon. The most powerful instance on offer includes eight chips from Nvidia’s A100 graphics card series, which was the company’s flagship AI chip line before the introduction of the H100 last year. 

Paperspace offers two kinds of data storage alongside its compute infrastructure. Developers can provision dedicated block storage for each instance, as well as create so-called shared drives. The latter feature enables multiple instances to access the same data repository, which eases tasks such as exchanging data between AI models.

After companies develop a custom neural network, they usually productize by integrating it into their applications. One common way to go about the task is to deploy the neural network on cloud infrastructure and make it accessible through an application programming interface, or API. Workloads can send tasks to the AI model via the API and receive processing results the same way.

Paperspace promises to ease the API creation workflow. According to the company, its platform enables developers to make a neural network accessible via a programming interface in only a few steps. It achieves that automating many of the manual infrastructure-level tasks usually involved in the process. 

In the wake of the acquisition, DigitalOcean plans to integrate Paperspace into its cloud platform. According to the company, the startup’s technology will make it easier for its customers to build and deploy AI models. The deal will also create opportunities for DigitalOcean to cross-sell its existing services to Paperspace’s users.

“The combined offerings allow customers to focus more on building applications and growing their businesses and less on the infrastructure powering them,” said DigitalOcean Chief Executive Officer Yancey Spruill.

DigitalOcean doesn’t expect the deal to boost its financial results meaningfully this year. However, the company estimates that Paperwork’s platform will start accelerating its revenue growth from 2024 onward. 

Image: Unsplash